The present invention relates generally to the field of data processing systems, and more particularly to computing composite functions in a map-reduce framework.
Parallel processing is used to increase the speed of execution, and amounts of data to be processed. Map-reduce systems provide a framework for parallelizing processing tasks to be performed on large data sets. In a map-reduce framework, mappers process data chunks assigned to them and send their results to reducers, which each receive a portion of all mapper results belonging to a certain subset of a task.